I was struggling with sleep, so I started chatting with Claude—just to think through the problem. The idea was simple: get some basic suggestions, try a few things, crash for the night, then track patterns over time and see what might be going wrong.

At some point, almost casually, I asked: should I just build my own sleep-tracking app?

The tool said no, and gave me some genuinely useful advice instead. But what struck me wasn’t the answer. It was the fact that the question even made sense to ask. Not that I could build it—as surprising as that might be—but that having this thought in the first place is now a logical thing. The mere possibility of thinking this way blows my mind.

That we live in a world where an ordinary person can seriously consider building a small, personal piece of software—tailored exactly to their own quirks, habits, and needs—is kind of shocking.

Think about it. On November 29, 2022—the day before ChatGPT launched—99% of non-technical people would have struggled to build a custom blog. I’m not saying they couldn’t build a blog. They had WordPress, Substack, Ghost, whatever. But if you wanted to even moderately customize it beyond off-the-shelf themes and changing some colors? You couldn’t do it.

Fast forward to 2025, and people are spinning up weird and wonderful and wacky-looking blogs and websites and apps using all sorts of frameworks they have absolutely no business even knowing: Node, React, SvelteKit, Astro, Hugo, Zola. I still don’t know Node and React from bhaaji puri and stuffed paneer paratha with aloo and mutter. And yet here we are, using them.

This happened in three years. The fact that you can conjure things into existence—make whatever you’re imagining come true—just by using plain English instructions. That’s magical. I know I’ve said similar things in my last few posts, but I keep coming back to it because we’re always poor at grasping the sheer quantum of change, especially something like AI.

Here’s another example of the magic. This very post you’re reading? I was sitting with my morning filter coffee. A fleeting thought struck me. I opened ChatGPT voice mode, dictated my raw thoughts, got a transcript. Dropped it into Claude, asked it to clean up the filler and tighten the structure. It gave me feedback, pointed out some interesting angles I hadn’t considered. I added more thoughts, it cleaned those up too. And now it’s live here. Pretty much everything I write on this blog has been voice-written. It’s helping me unlock a new and very expressive way of writing compared to my long-form pieces, where I actually have to commit the disgusting act of using a keyboard and pressing those god-awful buttons.

Now, this isn’t to say that all such tools will be good, or useful, or even work. I don’t think those are the right frames. Plenty of this will end up as disposable software: half-finished experiments, abandoned scripts, little digital detritus. I’ve written before about this in a post called Code Is Now Content. That landfill is inevitable. And yes, there are downsides—but that’s a different topic altogether.

What matters is this: people now have access to something that feels like a magic genie. A system that can help you conjure tools on demand, just to solve your problem, in your context, for your life. That is an extraordinary shift.

As Stewart Brand put it:

“We are as gods and might as well get good at it.”

We really are living in wild times.